The purpose of this study was to identify the characteristics of groups that construct understanding among learners in cooperative learning. As a result of the investigation, it was found that deeper understanding groups repeatedly complemented each other’s utterances in the first half of the interaction process. After that, follow-up joint caution utterances were repeated, and it was shown that understanding was deepened by the occurrence of adjusted utterances. On the other hand, in groups which lacked joint attention, the learners focused on explanations rather than waiting for others to share their attention. Therefore, the act of following up on the subject of interest together was interrupted, and no adjusted utterances occurred. In other words, in cooperative learning, it is important for learners to continuously follow their attention until the adjusted utterances that construct the learner’s understanding appear. It was also suggested that follow-up joint attention behavior occurs between learners in an equivalent relationship, not in a guided relationship.